A total of 90 male SHR (aged 10–12 weeks, weight, 250–280 g) were obtained from the Experimental Animal Center of Sun Yat-sen University (Guangzhou, China). All the animal experiments performed in this study were approved by the Animal Ethics Guidelines for the Care and Use of Laboratory Animals of the National Institute of Health (Publication No. 80-23, revised 1996).

SHR were randomly divided into four groups (n = 20 rats/group): (1) a Sham group that received sham surgery; (2) a tMCAO (2 d) group, in which rats underwent tMCAO surgery and NOR tests were performed 2 days later; (3) a tMCAO (28 d) group, in which NOR tests were performed at 28 days post-tMCAO; and (4) a tMCAO (28 d) + PE group, in which tMCAO was established and rats exercised in a running wheel for 26 consecutive days starting at the third day post-tMCAO. Besides, 10 rats were received sham surgery and exercised in a running wheel for 26 consecutive days starting at the third day post-surgery, as a Sham + PE group.

SHR were anesthetized by intraperitoneal (IP) injection of chloralhydrate (350 mg/kg), and tMCAO was then performed as previously described (Zhang et al., 2013). In brief, a 4-0 nylon suture filament with a rounded tip was inserted into the internal carotid artery to block the middle cerebral artery at its origin. After 90 min, the filament was removed to allow reperfusion. Body temperature was maintained at 37 ± 0.5°C by a heat lamp (FHC, Bowdoinham, ME, USA) throughout the surgery. The rats in the Sham group underwent sham surgery without suture insertion.

Rats subjected to running wheel exercise plan were placed into a programmable, motorized wheel apparatus (21 cm diameter, 40 cm length; South China University of Technology, Guangzhou, China), permitting the quantification of exercise intensity. Rats in the PE group were placed into the wheel and subjected to a period of exercise for 26 days starting at day 3 post-tMCAO. During the first 12 days, the running speed was set at 5 rev/min (approximately 3 m/min), for 20 min, twice a day (8 a.m. and 8 p.m.). For the next 14 days, the running speed was increased to 10 rev/min (approximately 6 m/min). The rats in the Sham and tMCAO groups were housed in a standard cage with free access to food and water, with no specific exercise plan.

The infarct volume of SHR in each group (n = 5) was assessed using 2,3,5-Triphenyltetrazolium chloride (TTC) staining (Wang et al., 2016). Briefly, the brains of SHR from Bregma +4.0 mm to −6.0 mm were sliced into six 2.0-mm-thick sections. After incubation in 2% TTC solution (Sigma, St. Louis, MO, USA) for 30 min at 37°C, coronal brain sections were then fixed in 4% paraformaldehyde solution overnight. The infarct area was identified by non-staining region while the live area should turn red. Image analysis was carried out independently by two observers (both XP and LZ with 4 years of experiences). The volume of infarction was measured with Image-Pro Plus software (Media Cybernetics, Silver Spring, MD, USA). The relative infarct volume was manually calculated according to the following formula: infarct percentage = (volume of the contralateral hemisphere − volume of the non-infarct contralateral hemisphere)/volume of the contralateral hemisphere × 100%.

The modified neurological severity scores (mNSS) was recorded at 2, 14 and 28 days post-tMCAO as described previously (Liu et al., 2011). The mNSS is a composite of the motor, sensory, balance and reflex tests and is assessed on a scale of 0–18 (normal score 0, maximal deficit score 18). Rats were tested three times by an individual blinded to grouping, and the average was recorded.

NOR tests were conducted in an open box (72 × 72 × 35 cm) with clear plastic walls and a black plastic floor. The box was placed in a dark room under infrared light throughout the test. On the day before the NOR test, rats were allowed to explore the empty box freely for 1 h. In the first trial, two identical objects (A and B) were placed in the upper right and lower left quadrants of the box, and placed rats into the box. After 10 min of exploration, rats were removed from the box and allowed a 1 h break. In the second trial, object B was replaced with object C (with C being dissimilar to both A and B). The rats were then put back into the box for a further 5 min period of exploration. The behavior of rats was video recorded and the evaluation of NOR memory was expressed as a percentage of the discrimination ratio calculated according to the following formula: Discrimination ratio (%) = (N − F)/(N + F) × 100%, where N represents the time spent in exploring the new object and F represents the time spent in exploring the same object.

On days 2 and 28 post-tMCAO, rats (n = 10) were anesthetized with 1 g/kg chloralhydrate and perfused transcardially with 150 mL normal saline (4°C), followed by 200 mL of 4% paraformaldehyde (4°C; Zheng et al., 2014). The rats from sham group (n = 10) were subjected to the same steps. The brains were then removed and fixed in 4% paraformaldehyde for 8 h at 4°C, prior to sequential immersion in 20% and 30% sucrose until the tissue no longer floated in the solution. Coronal sections were cut on a cryostat (CM1900; Leica, Heidelberger, Germany). The sections of the prefrontal cortex from the Bregma +4.20 to +2.20, the corpus callosum from the Bregma 0.20 mm to −2.20 mm and the entorhinal cortex from the Bregma −4.80 to −7.04 were processed for immunofluorescence staining. Coronal sections (30 μm thick) were processed for Luxol Fast Blue (LFB) staining.

Immunofluorescence staining of NeuN, Nestin, Ki67, MBP, SMI-32, SYN and PSD-95 was performed using previously reported methods (Zheng et al., 2014). Briefly, sections were pre-treated with citrate buffer (0.01 mol/L, pH 6.0) for 5 min at 85°C, and then incubated in 5% normal goat serum for 1 h at room temperature. Subsequently, the sections were incubated overnight at 4°C with antibodies for the detection of mouse anti-NeuN (1:200; Millipore, Billerica, MA, USA), mouse anti-Nestin (1:200; Millipore), rabbit anti-Ki67 (1:500; Abcam, Cambridge, MA, USA), rabbit anti-MBP (1:500, Abcam), mouse anti-SMI-32 (1:500, Calbiochem, NE1023), mouse anti-SYN (1:100; Abcam) and rabbit anti-PSD-95 (1:1000, Abcam). After washing three times with phosphate-buffered saline (PBS), the sections were incubated with the appropriate peroxidase-conjugated secondary antibodies anti-mouse IgG (1:1000, Cell Signaling Technology, Danvers, MA, USA) and anti-rabbit IgG (1:1000, Cell Signaling Technology) for 1 h at room temperature. Fluorescence signals were observed with an optical microscope (BX51; Olympus, Tokyo, Japan).

All histological images captured using the same exposure conditions were analyzed with the Image-Pro Plus analysis software (Media Cybernetics, Silver Spring, MD, USA). Eight consecutive sections of the prefrontal cortex, corpus callosum and entorhinal cortex were analyzed. The cell counts in four non-overlapping fields (425 μm × 320 μm) under an optical magnification of 400× were recorded and presented as the average cell number per field for each section.

The sections were first immersed in xylene followed by anhydrous ethanol gradient dehydration. Thereafter, the sections were stained overnight in 0.1% LFB solution (Sigma-Aldrich, St. Louis, MO, USA) at 60°C. After washing with 95% ethyl alcohol and distilled water, the sections were differentiated in 70% ethyl alcohol for 30 s. Differentiation was terminated by washing in distilled water until the unmyelinated tissue appeared white. For Nissl staining, the sections were stained with cresyl violet to assess changes in the number of Nissl bodies.

Deoxynucleotidyl Transferase-Mediated dUTP in situ Nick-End Labeling (TUNEL) assay was performed in the entorhinal cortex region, using an in situ cell death detection kit (Fluorescein; Roche Corp., Basel, Switzerland). TUNEL signals were detected with an optical microscope (BX51; Olympus). Visible TUNEL-positive cells were counted by Image-Pro Plus image analysis software in each group.

The remaining SHR (n = 5) were deep narcotized in each group and perfused intracardially with 50 mL ice-cold 0.9% saline. Then the entorhinal cortex of brain tissue (from Bregma −4.80 to −7.04) was rapidly dissected and homogenized in protein extraction buffer (Thermo, Pierce Biotechnology, Waltham, MA, USA) containing a complete protease inhibitor cocktail (Thermo). Protein concentration was measured with the Bradford Protein Assay (Thermo). Equal amounts of the protein samples were separated by 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred onto polyvinylidene difluoride (PVDF) membranes. After blocking with 5% non-fat dried milk for 1 h at room temperature, the membranes were incubated with rabbit anti-nestin (1:200; Millipore), mouse anti-synaptophysin (1:100, Abcam) rabbit anti-Bcl-2 (1:1000; Cell Signaling Technology) and mouse anti-β-actin (1:1000; Santa Cruz) overnight at 4°C. The membranes were then incubated with the appropriate secondary antibodies (1:1000; Cell Signaling Technology) for 2 h at room temperature. Positive signals were detected by enhanced chemiluminescence (ECL, Cell Signaling Technology) and visualized by exposure to X-ray film.

The results were presented as the mean ± SEM. Statistical analysis was performed by using SPSS version 13.0 (SPSS Inc., Chicago, IL, USA). Statistical differences of mNSS in different groups were analyzed by the repeated measures ANOVA. All other statistical differences were analyzed by ANOVA with a LSD-t procedure. P-values < 0.05 were considered to indicate statistical significance.

Previous research has emphasized that tMCAO induces brain infarction and neurobehavioral function impairment (Lourbopoulos et al., 2010; Wang et al., 2015); this was also confirmed in our study (Figure 1). Obvious brain infarcts were observed 2 days post-tMCAO. Subsequently, the infarct volumes decreased, with minimum infarct volumes detected at day 28 post-tMCAO (P < 0.001, Figures 1B,C). Rats at day 2 post-tMCAO showed an extremely high mNSS (8.4 ± 1.0), and the score was decreased to 5.4 ± 1.6 at day 14 and to 3.1 ± 0.9 at day 28 (Figure 1D). The repeated measures ANOVA revealed a significant main effect of (mNSS of tMCAO + PE group < mNSS of tMCAO group, F_(1,9) = 6.076, P = 0.024), an significant interaction between treatment effects and time effects (F_(2,36) = 16.051, P < 0.001) and a significant time effect (F_(2,36) = 430.122, P < 0.001). Of note, rats in the tMCAO + PE group showed significant decreases in infarct volumes and mNSS when compared with rats in the tMCAO group without PE (P < 0.05 and P < 0.01, respectively), which indicated that PE is beneficial in neurological function recovery after brain infarction.

Next, we investigated the ability of PE to improve the recovery of NOR memory in tMCAO rats. No significant difference of discrimination ratio was observed between Sham and Sham + PE groups (P > 0.05, Supplementary Figure S1). As shown in Figures 1E,F, the discrimination ratio was decreased at day 2 post-tMCAO, and reached a minimum 28 days later (F = 27.39, P < 0.001). At day 28 post-tMCAO, rats subjected to PE exhibited a much higher discrimination ratio than that of rats subjected to tMCAO without PE (26.6 ± 16.9% vs. −31.0 ± 28.4%; P < 0.01). These data indicated that ischemic stroke was related with the decline of NOR memory and PE improves NOR memory in tMCAO rats.

Increasing evidence indicates that the prefrontal and entorhinal cortex are closely involved in the neural circuitry of NOR memory (Wilson et al., 2013; Chao et al., 2016). Thus, we examined the neurons in the prefrontal cortex and the entorhinal cortex after stroke to evaluate the impacts of PE on neurogenesis in ischemia-injured brains. In the prefrontal cortex, the NeuN-positive cell numbers in the Sham, tMCAO and tMCAO + PE groups were equivalent (F = 1.22, P = 0.32, Figures 2A,B). However, the NeuN-positive cell numbers were significantly reduced in the entorhinal cortex post-tMCAO (P < 0.001), with minimum numbers observed at day 28 post-tMCAO (P < 0.05; Figures 2A,C). The number of NeuN-positive cells in rats subjected to PE was much higher than that in the tMCAO rats without PE (22.1 ± 5.3 vs. 4.7 ± 2.4; P < 0.001).

In addition to the prefrontal and entorhinal cortex, the corpus callosum has also been reported to be associated with mild cognitive impairment (Blasi et al., 2014; Brown et al., 2016). Therefore, we performed immunofluorescence staining of MBP (a marker of myelin integrity) and SMI-32 (a marker of abnormally dephosphorylated neurofilament protein) to further evaluate pathological changes in the corpus callosum, as well as LFB staining (a method used to stain myelin and phospholipids) in the corpus callosum. As shown in Figures 3A–D, there were no remarkable differences in the expression of MBP (F = 1.938, P = 0.143) or SMI-32 (F = 0.074, P = 0.974) between the Sham, tMCAO and PE + tMCAO groups. Furthermore, no obvious changes in LFB stained myelin and phospholipids were observed in rats post-tMCAO.

Nissl staining and immunofluorescence staining of NeuN were performed to confirm whether PE recovered ischemic stroke-induced neuronal injury in the entrohinal cortex. As results shown in Supplementary Figure S2, PE slightly increased the numbers of Nissl-stained cells and NeuN-positive cells after sham surgery, and these increases did not reach significant difference (P > 0.05). tMCAO groups (2 or 28 d) exhibited decreases in the numbers of Nissl-stained cells and NeuN-positive cells compared with Sham group. As expect, the numbers of Nissl-stained cells and NeuN-positive cells were remarkably increased in tMCAO + PE group, when compared with tMCAO groups (P < 0.01 or P < 0.001). To further investigate the therapeutic mechanism of PE, we detected the expression of proliferation markers (Nestin and Ki-67) and apoptosis index (TUNEL) in the entorhinal cortex after ischemic stroke (Scholzen and Gerdes, 2000; Endesfelder et al., 2017). As shown in Figures 4A–C, Nestin-positive cell numbers were significantly increased at day 28 post-tMCAO (P < 0.05). Ki-67-positive cell numbers were also slightly increased after tMCAO, although this effect did not reach the level of statistical significance. PE dramatically upregulated the expression of Nestin and Ki-67 compared with that in the tMCAO and Sham groups (P < 0.01 and P < 0.001, respectively). Besides, the number of TUNEL-positive cells was significantly increased after tMCAO (P < 0.001) and then restrained by PE (P < 0.001; Figures 4A,D).

The expression levels of MBP and SMI-32, and two synaptic plasticity-associated proteins (SYN and PSD-95) were also detected in entorhinal cortex. As shown in Figures 5, 6, immunofluorescence staining revealed increased expression of SMI-32 after tMCAO (P < 0.001), while expression was decreased after PE intervention (P < 0.001). Conversely, expression of MBP, SYN and PSD-95 were all decreased after tMCAO (P < 0.001), while expression of all three markers was increased after PE (P < 0.01).

Western blotting analysis was performed to assess the expression of Nestin, Bcl-2 and SYN proteins in the entorhinal cortex. As shown in Figure 6D, the western blot results were consistent with those obtained by immunofluorescence staining. The expressions of Nestin and Bcl-2 proteins were remarkably up-regulated by PE after stroke. The expression of SYN was down-regulated in post-tMCAO, but was also up-regulated after PE intervention.